PINDAR S LIBRARY: PERFORMANCE POETRY AND MATERIAL TEXTS

PINDAR S LIBRARY: PERFORMANCE POETRY AND MATERIAL TEXTS TOM PHILLIPS DPhil THESIS WORD COUNT: 99, 863

ACKNOWLEDGEMENTS It is a great pleasure to acknowledge the many debts I have incurred before and during the writing of this thesis. My Classics teachers at Monmouth School, Peter Dennis Jones and David Jenkins, and later Ben Giles and David Hope, first fostered my interest in the classical languages. During my time at Oxford I have been fortunate to share the company of numerous scholars who have informed and changed the way I approach literature; particular thanks are due to Felix Budelmann, Peta Fowler, Stephen Harrison, Gregory Hutchinson, Francesca Martelli, Dirk Obbink, Scott Scullion, Guy Westwood and Tim Whitmarsh. Oliver Thomas has read much of what I have written over the last few years, and given much exacting advice. My doctorate was undertaken with the support of funding from the AHRC, for whose assistance I am very grateful. My greatest obligations, personal and academic, are to Tim Rood, my undergraduate tutor, and my doctoral supervisor, Armand D Angour, who have both provided advice, criticism, and encouragement well beyond the call of duty. Corpus Christi College, Oxford December 2012 T.R.P 2

Contents Abbreviations 4 Introduction 6 PART I. CONTEXTS: TO ALEXANDRIA AND BEYOND 1. Writing In The Library 45 2. Texts and Metatexts 63 3. Passing on the Garland: Reception and Material Sites 82 PART II. POEMS ON THE PAGE 4. Edited Highlights 110 5. Epinician Archives: Olympians 10 and 13 147 6. Closing the Book: Olympian 14 182 7. Pythians 11 and 12: Materiality, Intertextuality, Closure 203 Conclusions 236 Bibliography 240 3

ABBREVIATIONS AP Chantraine Anthologia Palatina. P. Chantraine, Dictionnaire étymologique de la langue grecque: histoire des mots (Paris, 1968). Cults R. Farnell, The Cults of the Greek States, I V (Oxford, 1896 1909). DFA D-K A. Pickard Cambridge, The Dramatic Festivals of Athens, 2 nd edition revised by J. Gould and D. Lewis (Oxford, 1988). H. Diels and W. Kranz eds., Die Fragmente der Vorsokratiker, I III (Berlin, 1974). EG D.L. Page ed., Epigrammata Graeca (Oxford, 1975). FGrH F. Jacoby et al. eds., Fragmente der Griechischen Historiker (Leiden, 1923 ). Fowler R.L. Fowler ed., Early Greek Mythography I: Texts (Oxford, 2000). GP Hummel LIMC A.S.F. Gow and D.L. Page eds., The Greek Anthology: Hellenistic Epigrams (Cambridge, 1965). P. Hummel, L épithète pindarique: étude historique et philologique (Bern, 1999). Lexicon Iconographicum Mythologiae Classicae, I VIII (Zurich/Munich, 1981 99) K-A R. Kassel and C. Austin eds., Poetae Comici Graeci (Berlin, 1983 ). LfgrE Lexikon des frühgriechischen Epos (Göttingen, 1979 ). LSJ Müller H.G. Liddell, R.Scott, H.S. Jones, and R.Mackenzie eds., A Greek English Lexicon (9 th edition, Oxford, 1940). C. Müller, Geographi Graeci Minores (Paris, 1855 82 [reprinted Hildesheim, 1965]). M-W R. Merkelbach and M.L. West eds., Fragmenta Hesiodea (Oxford, 1967). PEG A Bernabé ed., Poetarum epicorum Graecorum testimonia et fragmenta I (Leipzig, 1987). Pf. R. Pfeiffer ed., Callimachus, I II (Oxford, 1949, 1953). Pitiche Pindaro Le Pitiche: introduzione, testo critico e traduzione di Bruno Gentili, commento a cura di P. Bernardini, E. Cingano, and P. Giannini (Milan, 1995). 4

PMG D.L. Page ed., Poetae Melici Graeci (Oxford, 1962). POxy The Oxyrhynchus Papyri, London 1898. Race SH SIG W.H. Race ed. Pindar, Olympian, Pythian, Nemean, Isthmian Odes, I II (Cambridge, Mass., 1997). Supplementum Hellenisticum, eds. H. Lloyd-Jones and P. Parsons (Berlin/New York, 1983). Sylloge Inscriptionum Graecarum, I IV ed. W. Dittenberger (3 rd ed., Leipzig, 1915 24). Slater W.J. Slater, Lexicon to Pindar (Berlin, 1969). SLG D.L. Page ed., Supplementum Lyricis Graecis (Oxford, 1974). S-M B. Snell and H. Maehler eds., Pindari Carmina cum Fragmentis (Lepizig, 1984, 1989). TGrF R. Kannicht, S. Radt, and B. Snell eds., Tragicorum Graecorum Fragmenta, I- (Göttingen, 1971 2004). Wendel C. Wendel ed., Scholia in Apollonium Rhodium Vetera (Berlin, 1935). 5

INTRODUCTION τὸν Ὀλυµπιονίκαν ἀνάγνωτέ µοι Ἀρχεστράτου παῖδα, πόθι φρενός ἐµᾶς γέγραπται γλυκὺ γὰρ αὐτῷ µέλος ὀφείλων ἐπιλέλαθ ὦ Μοῖσ, ἀλλὰ σὺ καὶ θυγάτηρ Ἀλάθεια Διός, ὀρθᾷ χερί ἐρύκετον ψευδέων ἐνιπὰν ἀλιτόξενον. Read out to me the name of the son of Archestratus, the Olympian victor, where it has been inscribed in my mind, for I have forgotten the sweet song owed to him. O Muse, and you Truth daughter of Zeus, with straight hand ward off the reproach of lying and wronging a friend. So begins Pindar s ode for Hagesidamus of Western Locri, celebrating his victory in the Olympian boys boxing competition of 476 BC. This is one of several uses of the metaphor of writing in Pindar. At N.6.6-7, Pindar employs inscription to describe the activity of fate in setting down a marker for human life, of which mortals are unaware (καίπερ ἐφαµερίαν οὐκ εἰδότες οὐδὲ µετὰ νύκτας ἄµµε πότµος / ἅντιν ἔγραψε δραµεῖν ποτὶ στάθµαν). Here, as at O.10.1-3, the act of inscription gives rise to a substrate the permanence of which contrasts with the contingencies of human mental activity. A somewhat different and more diffuse focus on writing emerges at O.6.90-1, where the narrator addresses Aeneas, the chorus-trainer, as a message-stick of the Muses (ἐσσὶ γὰρ ἄγγελος ὀρθός, / ἠϋκόµων σκυτάλα Μοισᾶν, γλυκὺς κρατὴρ ἀγαφθέγκτων ἀοιδᾶν); Aeneas responsibilities presumably involved taking a physical copy of the poem to the place of performance and instructing the chorus. 1 The phrase evokes the Spartan method of sending written messages. The σκυτάλη was a staff around which a strip of leather or papyrus would be wound; one was kept in Sparta and the other, of exactly the same size, was given to, for example, the commander of an army. The papyrus would be wound slantwise around the staff and the message written vertically, 1 Cf. Hutchinson (2001) 417. 6

so that it could only be read correctly when wrapped around its counterpart. 2 Pindar s metaphor draws on this custom, casting poetic performance as a participation in a shared body of specialist knowledge; like the Spartan users of the σκυτάλη, Aeneas, and by extension the performing chorus, have sufficient skill to be able to understand the poem and translate it into a performance. 3 The passage uses the separation of a poetic utterance from its author as an opportunity of celebrating the communality of performance culture and, by implication, writing s power to fix and disseminate texts. 4 These passages form part of a more general pattern in Pindar s self-referential rhetoric, in which songs are characterized as objects, particularly those which commemorate athletic success such as crowns, fillets, and statues. 5 These parallels often have an agonistic dimension, as at P.6.10-14 where the poem qua ὕµνων θησαυρός is not vulnerable to the elements in the same way as a real building, or N.5.1-5 where song s mobility is contrasted with the fixity of statues. These and other similar passages can be read as insisting on the pragmatic superiority of song to other artforms, 6 but they also contribute to a metaphorization of the text itself as an object involved in a nexus of social relationships. 7 In addition to their particular contextual functions, these passages can be seen as responses to and manipulations of the technological changes taking place in Pindar s lifetime. The development of writing during the archaic period is difficult to trace precisely, but it is clear that by the time of Pindar s poetic career writing was an important part of Greek culture and education; contemporary vase paintings testify to the increasing use of various types of writing materials in teaching, 8 and the Muses are shown carrying papyrus rolls, signalling the connection between writing and poetic composition. Metaphors involving writing become increasingly common from around the 470s, 9 and although the first half of the fifth century 2 This is the explanation of its function given by Σ O.6.154b and f, although a slightly different account is given by d. Cf. e.g. Thuc. 1.131; Xen. Hist. 3.3.8; Ar. Lys. 991; Plut. Lys. 19. 3 For the musical education and understanding which composers and performers brought to bear on the melodization of poems during this period cf. D Angour (2007). 4 Cf. Hutchinson (2001) 415. Herington (1985) 45 argues that in the archaic and classical periods texts were no part of the performed poem as such, but merely a mechanical means of preserving its wording between performances. This, however, mistakenly discounts the possibility of readerly engagement with texts. Hubbard (1985) 67 n. 166 thinks that Pindar would have held written dissemination to be even more important than performance: this inference is unjustified, and oversimplifies the relationship between the two modes. 5 Cf. Steiner (1994) 92 nn. 107-9. 6 Cf. Svenbro (1976) 172-93; Steiner (1994) 95-6; Thomas (2007); Kowalzig (2011). 7 Thus Steiner (1994) 96-9. N.b. particularly her comments p. 96: [t]hrough the presence of the monuments, the poet suggests a resolution to two issues central to his craft: first, his own relation to his patron, second the question of the afterlife of his compositions, their ability to go on sounding the athletes praise. 8 Immerwahr (1964) and (1973) provide an inventory of depictions of books on vases. 9 Cf. e.g. Aesch. Supp. 179; PV 459-62; TGrF F 597. For the connection between these passages and the appearance during this period of school scenes on vases see Ford (2003) 24. Hdt. 6.27 tells a story about 7

cannot be described as fully literate in the manner of, say, the Hellenistic period, 10 writing was certainly playing a role in the creative process. Equally, writing influenced other aspects of performance culture; reperformance of epinician odes, for example, probably depended heavily on written texts, and there is some evidence for their preservative function. One such datum is the famous story about Pindar s epinician for Diagoras of Rhodes, O.7, being inscribed in the temple of Lindian Athena (ταύτην τὴν ᾠδὴν ἀνακεῖσθαί φησι Γόργων ἐν τῷ τῆς Λινδίας Ἀθηναίας ἱερῷ χρυσοῖς γράµµασιν, Gorgo says that this ode was dedicated in the temple of Lindian Athena in golden letters, Σ O.7 inscr. Drach. I p. 195) which likely represents wider practice; while the golden letters are probably apocryphal, the inscription of such poems, or the dedication of written texts in temples and private archives, could well have been widespread. 11 Despite these hints, however, our understanding of how written texts of Pindar operated in the 5 th and 4 th centuries is severely limited by lack of evidence. No papyri survive, and we are in the dark about many of the precise details of textual production; how exactly Pindar made use of writing in the process of composition (or if he did at all although this seems highly likely), 12 what the texts of his poems looked like, how and how widely they circulated, who read them, 13 and the precise details of how they served as bases for reperformance are all largely matters of guesswork. What Pindaric books were like before the Hellenistic period, whether they consisted of single poems or small groups, or of larger collections similar to those of later scholarly editions, is equally obscure. 14 Nevertheless, various critics have given children in Chios being taught writing, an episode which occurs in 496. The evidence for written texts pre-431 is collected by Herington (1985) 201-6. 10 Although the opposition between oral and literate is itself problematic, as numerous studies have shown: cf. e.g. Duguid (1996) 66-73. See Scodel (2001) 125 on the intersection of orality and literacy in Pindar. 11 On the possible dedication of a copy of O.7. by Diagoras cf. Carey (2007) 201 n. 13. For the vocabulary of dedication cf. Herington (1985) 46; Rutherford (2001) 175-8. Writing is also used to describe an act of dedication at O.3.30. 12 Cf. Herington (1985) 46. 13 The evidence for the book trade in the classical comes mainly from the late fifth and fourth centuries: cf. Ar. Ran. 52-4; Xen. Anab. 7.5.14 and Harris (1989) 85 n. 92. Pl. Ap. 26d-e has Socrates mention that τὰ Ἀναξαγόρου βιβλία are available for a drachma in the agora. Currie (2004) 52 sites the development of the book trade in the late fifth century, but there is no reason to suppose that books were not circulating earlier than this, albeit perhaps on a smaller scale. Hubbard (2004) argues for the importance of written texts in the dissemination of epinician in the fifth century; Carey (2007) 200 n. 5 is more cautious. 14 Cf. Irigoin (1952) 5-28 for an assessment of the evidence. Particularly interesting are his arguments p. 16 that the allusions to Pindar in Aristophanes are based on an Athenian edition of a selection of the poems: [l]e choix de ces oeuvres correspond sans aucun doute au programme des oeuvres lyriques qu on enseignait aux jeunes Athénienes entre 450 et 420... et donc à une partie du texte édité alors à Athènes, and p. 21 that regional editions were assembled and edited into larger editions at intellectual centres such as Athens and Rhodes during the late 5 th and 4 th centuries, editions which served as a basis for Hellenistic scholarship. Both cases are certainly plausible, but the nature of the evidence compels caution. It could, however, be argued that the allusions in comedy (for which see e.g. Eupolis fr. 366, Ar. Ach. 637-9, Birds 926-45, Clouds 1355-62, Knights 1323, 1329) need not reflect the contents of a particular edition, but rather the poems which were, for reasons of perceived 8

accounts of the use of writing, both as medium and metaphor, by Pindar and other poets, and this subject will doubtless continue to be a matter of debate. My chief goal in this study, however, is to explore the function of written texts of Pindar in later antiquity. In part this approach aims to correct a disparity between scholarly focus and textual history. Pindaric scholarship has in the main been concerned to analyse the social and historical conditions which shaped the poems production and initial reception, and to interpret how the poems operated in their fifth century contexts of performance. While such approaches are obviously important, we should not forget that the vast majority of readers in antiquity encountered Pindaric texts in book form, or in various reperformance scenarios remote from Pindar s own time. Attempting to understand what was involved in such encounters entails questions about the form of books, the nature of reading practices, and the effects of changing literary and scholarly contexts on readers approaches to the Pindaric corpus. Yet we should be wary of schematic oppositions between performance and books or reading, and between the contexts of Pindar s time and those of later periods. As we have seen, Pindar was acutely aware of the material forms his poems could take, and performance continues to be important, both as a concept with which later authors and scholars reimagined the circumstances of the fifth century poetic economy and as a mode of realizing texts in later periods. Similarly, although Pindar s poems are rooted in specific enunciative situations and respond in particular ways to the demands of individual social and political contexts, they also make generalizing claims which are easily assimilable to a variety of reading contexts. Indeed, the very specificity of features such as deixis, addresses to historical individuals, and references to particular cultic or political situations allows in later contexts for an imaginative recreation of the poem as a performance piece, permitting the poem to impose its own mode of contextualization upon the context(s) in which it is being read. 15 Pindar s poetry both transcends and embodies historical contingency. Moreover, the textual dissemination of epinicians can also be be seen as actualizing the texts own rhetorical strategies. One of epinician s primary concerns is the construction and aesthetic quality or relevance to Athenian affairs, best known to the audience; this knowledge could have come from reperformance as well as reading. The comic poets may have had access to considerably more of Pindar s poems than their allusive practices would seem to suggest, and of course we cannot discount the possibility that Pindar was cited or alluded to in works which are no longer extant. The loss of so much of the texts of authors associated with the New Music perhaps also distorts the picture; we might expect to find Pindaric allusions in these authors, and potentially to a wider range of texts than those found in the comic poets. 15 It should be noted, however, that the poems have little to say about their performance locations, a fact which eases the transition into reperformance and written dissemination: cf. Carey (2007) 199. Deixis in Pindar has received considerable attention: cf. Felson (1999); Athanassaki (2004); Felson (2004); Martin (2004). 9

perpetuation of the victor s fame, and scholars have become increasingly sensitive in recent years to the Panhellenic nature of these claims; as well as composing for particular contexts, Pindar is also writing for the whole of the Greek world. 16 This study aims to think, among other things, about this fame as an historical actuality as well as a textual projection (the two are obviously interrelated), and about how Pindar s immortalization of a figure such as Hiero of Syracuse operates in relation to changing historical circumstances. I shall therefore present readings of various poems through the lens of fifth-century performance in addition to considering their operations in later contexts; the latter cannot be understood without the former, but I shall also try to show how consideration of interpretative issues central to assessment of Pindaric texts in later periods can enhance our understanding of how the poems worked in Pindar s own time. The opening of O.10 quoted above is a useful starting point for thinking about these issues. The relationship between narrator and victor is presented through the lens of aristocratic guest friendship; 17 the narrator is in danger of wronging his friend (ἀλιτόξενον) by his tardiness in delivering the poem, 18 and the poem is presented as owed (ὀφείλων) to the victor. 19 The metaphor continues in the antistrophe, where the narrator describes his deep debt which shames him (ἐµὸν καταίσχυνε βαθὺ χρέος, 8) and hopes that the interest (τόκος, 9) accrued by the debt will loosen the sharp reproach (λῦσαι ὀξεῖαν ἐπιµοµφὰν, 9) attendant on it. The metaphorical structure sites the acts of composition and performance in a network of ethical concepts, and in the context of performance hints at the debt owed by the wider community to the victor s achievements. This is balanced by the figuration of the narrator s mind as a substrate that is simultaneously preservative and open to the contingencies of human understanding (ἐπιλέλαθ ), in need of assistance from the Muse and Aletheia in order to fulfil its duty. The use of γέγραπται in the context of the debt metaphor is usually interpreted as suggesting a ledger; the victor s name has been written in the narrator s mind in the same way as a debtor records the amounts he owes. 20 But when sung by a chorus, γέγραπται also hints at the distinction between composition and performance, 16 On Pindar s panhellenism cf. e.g. Athanassaki (2011a) with further references. 17 Cf. Verdenius (1988) 57. 18 For the possible historical circumstances of this situation cf. Verdenius (1988) 56 who suggests that the delay was due to Pindar being occupied with the more important odes O.1, O.2, and O.3. Cf. Erbse (1970) 28 for the reading that Pindar is lightheartedly teasing Hagesidamus over his eagerness to have his praises sung. Nassen (1975) 223-4 sees the description of lateness as balancing historical circumstances and poetic exaggeration. 19 For the topos of the debt owed to victory cf. e.g. O.1.103, O.3.7. Cf. Hubbard (1985) 65-8 for some remarks on the role of writing in this passage. 20 Thus Verdenius (1988) 54. 10

and between the poet and the chorus. 21 Read in this way, the name having been inscribed on my mind works as a reference to the processes of composition and training the chorus, which doubtless involved (a) written text(s). This connotation playfully, and perhaps humorously, undermines the rhetoric of debt and tardiness by implying that the poem was composed some time before the performance, and hence that the χρέος is not quite as serious as the second stanza makes out. Hagesidamus has (literally) been on the poet s mind, and the choral enunciation of a victor s name previously written down replays in miniature the dynamic of composition and performance. This implication is reinforced by the mention of Hagesidamus (Ἀρχεστράτου παῖδα) preceding the rhetoric of debt; the debt has, in the sequential movement of the text, been repaid before it has been constructed. The implied distinction between poet and chorus also opens up another way of understanding πόθι φρενός / ἐµᾶς γέγραπται, namely as a reference to the fact that Hagesidamus name has been more generally inscribed on people s minds by the act of victory, the subsequent announcement of his name at Olympia, and the word-of-mouth spread of the news that doubtless followed his success. All these lines of interpretation would also have been available to a later reader engaging with a written copy of the poem, albeit that the distinction between poet and chorus would have taken place at the level of imaginative recreation rather than through the actualities of performance. But the situation of the poem in a book also opens up other meanings, particularly in view of the interaction between the language of writing and the book itself. The act of writing in the poet s mind (γέγραπται) is not identical with any individual act of inscription: each physical copying of the text would have replayed this division and reinforced the phrase s metaphorical force. Together with the description of the poem as a γλυκὺ µέλος, the reference to (a metaphorical) writing invokes the difference between performance and reading. Yet the use of γέγραπται also interacts with its material aspect; the metaphor of writing on the mind is converted by the act of inscription into an (almost) literal description of the book itself, or, from another angle, might be read as doubly metaphorical, metaphorizing the book as the poet s mental substrate. Similarly, there is a difference between the enunciative situation of ἀνάγνωτέ µοι in performance and reading. 21 For the debate over monodic v. choral performance of epinicians cf. Heath (1988) and Lefkowitz (1988), both arguing that monodic performance was the norm, with the counter argument of Carey (1989). Currie (2005) 16-18 is a useful overview. A consensus has grown in recent years that choral performance was common, but that performance is a category which subsumes considerable variations in venue and size: cf. e.g. Carey (2001); Hornblower (2004) 33-6, and D Alessio (1994) on the difficulties of distinguishing authorial and choral first persons. There is no good reason to think that O.10 was not performed at its premiére by a chorus. 11

In reading the phrase, the reader describes his own activity, but also addresses the text, whereas the narrator, realized by the chorus in performance, uses the imperative absolutely without directing it to anyone in particular. 22 The reenunciation of ἀνάγνωτέ µοι by the reader as an address to the text highlights another aspect of the book, namely its function in preserving a temporally prior performance and, more generally, as offering a means of access to the past. Unlike the narrator, who dramatizes his mind (albeit ironically) as a somewhat unstable substrate, the reader interacts with the secure material frame of the book. Whereas in performance, φρενός / ἐµᾶς γέγραπται gestured to the process of written composition preceding performance, in the situation of the book the phrase points up the book s status as a record of performance which has been written. 23 These lines illustrate Pindar s dramatic use of the notion of writing. When considered in the light of their place in a book, they also alert us to the additional meanings and resonances that his poems can take on as a result of interactions between the text and its medium. The selfreferential characteristics of the passage make it particularly significant in this respect, but in the course of this study I shall demonstrate that similar kinds of arguments can be made about numerous other aspects of Pindar s corpus. MATERIAL ISSUES Scholars have devoted considerable attention to the subject of the book in antiquity. Recent studies have focused on books as aesthetically significant objects expressive of their owners status, on their influence on modes of literary composition, and on how they were produced, sold, and read. 24 Less attention, however, has been given to how the forms of books interact with the texts they contain, and it is this interaction with which this study is mainly concerned. Central to my analysis is a conception of the book as an aesthetically marked object in its own right, not simply as a neutral purveyor of meanings. Following the lead of contemporary book theorists, I want to reexamine the traditional division between semantic meaning and material form which conceives the latter as an incidental adjunct to the former, 22 Verdenius (1988) 55. Hubbard (1985) 67 sees it as addressed to the audience, which is then put in the position of reading the present ode externalized and objectified into a written text. 23 Cf. Edmunds (2001) 79 [t]hese techniques of orally performed poetry, Pindaric and archaic, are already grammatological, a potential deferral of the voice of the poet or performer(s) to other, later, voice(s). Relevant also is Payne (2006), who argues that Pindar s gnomic maxims have a transhistorical as well as a (temporally and spatially) local force: [g]nomic lyric presupposes its own transhistorical reception by addressing abstract formulations to a universal subject created by its own pronominal structures (p. 182). 24 Cf. e.g. Houston (2009); White (2009). Lowrie (2009) focuses on how writing is represented in Latin poetry; cf. also Dupont (2009); Farrell (2009). 12

and emphasise instead the book s status as a part of the processes by which meaning is created and maintained. This approach runs counter to one of the primary features of Western aesthetic thinking, in which the marginalization of the book as a part of aesthetic processes has been symptomatic of a wider view, derived chiefly from Plato and Aristotle, which privileges the noetic aspects of aesthetic experience over the material. 25 The book theorists emphasis on the material particularity of the book and its cultural significance problematizes this distinction, and also highlights the importance of the wider intellectual and social contexts in which books are produced, disseminated, and read. In tracing the distinction between formalist and materialist tendencies in modern criticism, Peter McDonald highlights the role of scholars such as Jerome McGann and Donald McKenzie in questioning New Critical and poststructuralist reading protocols which abstracted the text from its material context: They insisted on seeing the text, not as an abstract linguistic form, but as a mediated material artifact, a redescription which, they urged, entailed a significant shift in our understanding of the scene of reading. If this scene was defined for close readers by their critical engagement with what we could call the transcendent text-type the free-floating, idealized verbal text written by the author, it was structured more immediately for materialist readers by their physical encounter with an immanent texttoken a particular material document produced by various cultural mediators (editors, publishers, printers, etc.) for specific markets. 26 McDonald s analysis highlights the way in which the act of material inscription is itself meaningful, not only because the material document reflects a variety of social and institutional factors attendant on its production which are themselves part of the ongoing shaping of meaning, 27 but also because it generates interplay, or potential interplay, between meaning and medium. On this reading, the space of the document itself becomes a part of literary form. This position also points towards the tension between the book as a (provisionally) fixed structure and the instability of textual meaning, informed by its multiple 25 On the role of written texts in fostering more abstract ways of thinking about texts during the fourth century BC see Ford (2003). On books more generally see also Hutchinson (2008). 26 McDonald (2003) 231. His analysis builds on that of Chartier (1995) 134, who insists that [r]eaders never confront abstract, idealized texts detached from any materiality, and opposes a mode of reading which attempts to take account of the material conditions of textuality to the phenomenologies of reading promulgated by e.g. Iser (1980) which eras[e] the concrete modality of the act of reading and characteriz[e] it by its effects, postulated as universals. 27 See e.g. Chartier (1995); McKenzie (1999); (2002). 13

intertextualities, its potential for indeterminacies and its capacity for dislocating generic and ideological frames, and also by the multiple contexts in which it operates. This tension has been central to book theory and responses to it. In the context of the theoretical debates of the 1980s, the documentalist approach to the material particularity of the book was sometimes mobilized to oppose the poststructuralist notion of textuality decoupled from authorial control, even as its insistence on the roles played by extra-authorial and extratextual forces paralleled the anti-platonizing positions of textualists such as Derrida, Barthes, and Foucault. 28 And yet the apparent opposition between the documentalist insistence on the centrality of the material document and the poststructuralist commitment to a thinking of textuality which dislocates the borders, the framing of texts, everything which should preserve their immanence and make possible an internal reading 29 is far from absolute. A central poststructuralist insight has been the potentially infinite variety of contexts and their consequences for textuality; as it is only possible for meaning to emerge in contexts, meaning can never said to be securely finalized. McDonald argues that the poststructuralist approach complements documentalist emphasis on the material text in giving a theoretical framework to deal with the contextual and material variety attendant on the processes of dissemination: [I]mmanence [does] not entail stability, since, even in material terms, there is no end to the process of dissemination. Proliferation, not fixity, is the norm as texts are successively put to new uses in new forms. This is not, it should be stressed, simply a reassertion of the scholarly editor s traditional insistence on textual variation. It is a matter of recognizing the volatility of material contexts and the unpredictability of readings. Produced and reproduced by new cultural mediators, in new contexts, and for new readers, the successive versions of texts represent unique episodes in the constitution of meaning. 30 Using this model to think about books in the ancient world requires some methodological adjustment. Contemporary book theorists pay considerable attention to bibliographical issues 28 See Derrida (1978) 20 on the theological simultaneity of the book, which he treats in that discussion as an illusory totality linked to a metaphysical conception of the connection between meaning and presence. Cf. also Derrida (1976) 18-26 on the links between the book as traditionally conceived and metaphysics: [t]he idea of the book is profoundly alien to the sense of writing. It is the encyclopaedic protection of logocentrism against the disruption of writing [and] against difference in general (p. 18). 29 Derrida (1990) 92. 30 McDonald (2003) 232. 14

and social analysis of book production in a way precluded for classicists by lack of detailed evidence. 31 However, McDonald s emphasis on contextual variation is a useful starting point for thinking about the cultural issues which affected the Pindaric corpus in different periods of antiquity, as the texts became sites for scholarly contestation and mediation by individual readers, and were opened up to the influences of literary receptions. Equally, the volatility of material contexts is an apt formulation of the physical variety of ancient books. In antiquity, each book was a physically unique object, written out by a reader or a professional scribe. These inscriptional processes were shaped by economic considerations, by the professionalism of individual scribes, and by changing fashions in the style of bookhands, which were themselves of aesthetic significance. 32 Books were consequently idiosyncratic, often produced for and tailored to the needs of specific consumers, as well as modified by individual readers additions of marginalia. One of the concerns of this study will be to establish a dialogue between these material features (actual or deduced) and the texts themselves. While the distinction between text-type and text-token is helpful in focusing attention on the materiality of the book, we should resist an overly schematic application of this opposition. Every act of reading produces mental signifieds, and hence involves a certain abstraction of the text from its material instantiation. Reading Pindar is particularly marked in this respect, in that the performance language, self-referential apparatuses, and historical situations of the Pindaric texts prompt the reader to precisely such an abstraction, and encourage a retrojection of the text into a temporally prior situation. This is a mode of reading additionally encouraged by the ancient literary critical discourses and literary receptions which emphasise the connection between poetic utterance and the historical figure of the author and the circumstances of composition. What I want to emphasise here is that this mode of reading, while important, is not a natural approach which should be privileged, but rather one means of reading among others, one which is constructed in advance (in a variety of ways) by the Pindaric corpus itself, and one which historically is always in the process of being renegotiated, and influenced by contextual factors. The above discussion of the text-type/texttoken distinction is employed to foreground the mediations of the reading process by particular material factors, and to provide a model for discussing the various kinds of interactions that occur between signified structures and their material frames. 31 E.g. McGann (1991); McKenzie (1999). Fundamental also is Genette (1997) on the role of paratexts (titles, prefaces etc.) in presenting literary works and affecting their consumption. 32 On the concept of deluxe editions cf. Johnson (2004) 155-60; for a typology of Pindaric bookrolls in later antiquity cf. Ucciardello (2012) 113-18. 15

The status of Pindar s texts as performance poems plays an important role in their material resituation. The temporally unique, impermanent form of performance contrasts with the permanence of the material frame, and the gathering and ordering of the poems in editions imposes on them a form of collectedness at variance with the temporally discrete status of performances of individual songs. But as mentioned above, we should be wary of hard-andfast oppositions in this respect; the materiality of the book is to a certain extent prefigured by Pindar s characterization of his poems as objects, and various poems advertise their connections to others in the corpus even as performance pieces. 33 Despite these continuities, however, there are also important distinctions between the situations of performance and reading, some imposed by material conditions, and others by the strategies of the texts themselves in their altered contexts. References to musical accompaniment, choral singing, and other specific aspects of the performance economy impose on the reader a vivid sense of the difference of (aspects of) the poem s self-projection from the situation in which he himself engages with it, while at the same time influencing the reader s conception of his own activity. Any account of the reception of a textual corpus in antiquity must take account of ancient reading practices and the various ways in which these were conceptualized by literary critics. It was long a commonplace that reading in antiquity meant reading aloud, even when the reader was alone. 34 This position has been challenged by numerous scholars who have pointed out that, although voiced reading was clearly common, there is also much evidence for silent reading, and we should not think that voiced reading was the norm. 35 Debates about 33 See my remarks below pp. 106-20 on relations between poems within an edition. On performance intertextuality cf. Morrison (2011) with further references. 34 Cf. e.g. Balogh (1927); Hendrickson (1929); Rohde (1963); Kenney (1982) 12; Porter (2010) 314-15, 353-4; for further references cf. Gavrilov (1997) 56-8. On Roman reading practices cf. e.g. Quinn (1982); Cavallo (1999). Johnson (2000) examines the social practices in which reading was embedded and argues that whether reading was voiced or silent was dependent on context; cf. further Johnson (2009) on the representation of reading in Aulus Gellius and Johnson (2010) on the social functions of literary reading in the High Empire. The importance of sound to the reading process is demonstrated by the fact that the verb used to denote reading, ἀναγιγνώσκειν, is often synonomized by ἀκούειν, a situation paralleled by the synonymity of legere and audire in Latin. But whether these synonyms are (partially or wholly) metaphorical is debated: on the meaning of ἀναγιγνώσκειν cf. Chantraine (1950), who argues that the primary meaning is recognizing the letters; Carson (1999) 83 suggests that it is more specifically the recognition of sounds that is at issue, followed by Porter (2010) 351 n. 216. Cf. also Steiner (1994) 16-29. 35 Knox (1968) argues against the view that silent reading was not viewed as abnormal even if it was less common than voiced reading. He cites passages such as Cic. Tusc. 5.116, which discusses how one can compensate for the loss hearing, and suggests that a deaf man may still take pleasure in poetry: deinde multo maiorem percipi posse legendis his quam audiendis uoluptatem. Gavrilov (1997) sees the locus classicus of the evidential debate, August. Con. 6.3.3-4, as evidence for the normality of silent reading, and makes the claim that silent reading was normal in fifth century BC Athens as well as in later periods; cf. also Burnyeat (1997). Parker (2009) offers an overview of the Roman evidence and argues for the prevalence of silent reading, contending that public readings at conuiuia or readings by lectores were preparatory, ancillary, or supplementary to private 16

which form of reading predominated are less important for this study than the variety of reading situations in which texts would have been realized; reading aloud to oneself, to a small group of friends, reading silently on one s own, and engaging in a more dramatic performance before a larger group of a previously memorized poem (or part of a poem) were all possible scenarios, and having slaves read aloud was also a common practice. 36 Therefore, while we should be cautious about thinking that reading was usually or predominantly voiced, we need to think about the interpretative consequences of voiced reading and how this practice may have affected how readers orientated themselves towards texts. We also need to bear in mind the effects of literary theoretical discourses about the voice, and the role of poets voices in literary receptions. 37 Given the prominence of the emphasis on the voice in these texts, it is reasonable to assume that even when reading silently, ancient readers would sometimes have conceived of the texts they were engaging with as the written voice of an author, or at least entertained the possibility, as we still do today, of so conceptualizing the text. The notion of writing as inscribed/written voice (ἐγγράµµατος φωνή) is common in ancient theoretical literature. 38 Speech as an emitting of breath (πνεῦµα) was transcribed into the written document, which then became a trace not only of meanings but of the physicality of enunciation: voice inscribed in letters is the culmination of the breath that is stored within us (ἐγγράµµατος δὲ φωνή ἐστιν ἀποτέλεσµα τοῦ ἐν ἡµῖν ἐντεθησαυρισµένου πνεύµατος, Σ Dion. Thrax 212.23-4 Hilgard). This physicality could be foregrounded as a central part of the reading experience: Arcesilaus, head of the Middle Academy in the mid-third century BC, is quoted by Diogenes Laertius as describing Pindar as terrific at filling out the voice, a power he associates with a more general verbal sufficiency (τόν τε Πίνδαρον ἔφασκε δεινὸν εἶναι φωνῆς ἐµπλῆσαι καὶ ὀνοµάτων καὶ ῥηµάτων εὐπορίαν παρασχεῖν, Diog. Laer. 4.31). Reading (aloud) is, on these terms, an excavation of the authorial voice which inheres in the text: [t]he discourse produced by the reader s voice is a reenactment of the writer s voice that was transcribed in the act of writing. 39 The emphasis on the physicality of the voice is part of a wider theoretical concern reading. For an assessment of the conceptual underpinings of the silent/voiced reading debate cf. Fowler (forthcoming) 1-16. 36 Cf. Johnson (2000). 37 See below pp. 77-86. 38 For examples further cf. Porter (2010) 350 n. 212. 39 Bakker (2005) 41. Cf. Porter (2010) 310-11 for a characterization of this critical standpoint: render[s] the voice audible no matter how many layers of writing, transmission, quotation, or time have intervened the 17

in antiquity with issues of aesthetic materiality. Crates of Mallos, for instance, held that the quality of verbal sound was the most important factor is judging the value of a poem, 40 and Philodemus On Poems provides numerous testimonies to similar views held by other Hellenistic literary theorists. On Poems 1 fr. 91 records that Pausimachus makes an argument for the valuation of poetry according to sound on the basis of how it is enunciated: its quality does not reside in composition, but fine or inferior verse arises for no other reason than because of the combination of sounds itself (δι [οὐδὲν] ἕτερον φαῦλον [ἢ καλὸν] γίνεσθαι ἢ δι αὐ [τὸ τὸ συ]ν ηχεῖσ[θ]αι). This view is contested by Philodemus when he asserts in fr. 184, countering Andromenides, another euphonist, that everyone s hearing or mind is pleased not without reason (παν[τὶ τῆς ἀκο]ῆς ἢ τῆς δια[νοίας οὐκ ἀλό]γως οἰκειου[µένης). For Philodemus, hearing is not an irrational process separable from rational mental processes. What matters most for eliciting readerly reaction, on Philodemus account, is διανοία, the noetic structure which underpins formal effects. 41 The propositions that the sound of words has a function distinct from their meaning, that sound could be pleasing to the hearer, and that sound could reinforce meaning, were widely accepted by Hellenistic literary theorists, and form the background to the stylistic analyses of Cicero and Dionysius of Halicarnassus. The precise nature of linguistic sound and its relations to semantic meaning, however, were contested issues. The materiality of language is also a problematic figure in these debates; not only is there the primary argument about whether poetry is to be valued according to sound or meaning, but the status of the sound of language itself is in question. Whereas φωνή is often conceptualized as the voice of the author, it can also be employed to mean sound, and the euphonist thinkers cited by Philodemus often refer to a materiality inherent in language itself rather than a personalized voice. 42 This distinction will be significant when considering, for instance, the ways in which literary receptions of Pindar depersonalize the poetic utterance by view that within all linguistic expressions, oral or written, lies buried a voice that animates them. Cf. further ibid. 351-3. 40 Cf. Asmis (1992); (1995b) 152. Cf. further Philod. On Poems 5 col. xxv 2 xxix 18 for Crates argument that sound quality is not determined by rules but inheres naturally in an arrangement of words. Crates account of sound is not irrationalizing, however; he argues that one ought to judge through the senses the rational propositions present by nature [in the sound structures] (col. xxviii 19-33); for explication of this position cf. Jensen (1923) 146-73; Armstrong (1995) 265 n. 47; Asmis (2004) 12-13. 41 Cf. On Poems 5 col. xxxii 6-10, where Philodemus says that an unnamed opponent speaks only about lexis, while leaving aside thoughts, which have the far greater importance (καὶ περὶ τῆ[ς] λέ ξεως µόνον λαλεῖ, τὰ νο ήµατα κυριωτέραν δύ ναµιν ἔχοντα παραπέµ πουσ[α]), and PHerc. 1676 col. xiv (iii) 24-6 where Philodemus argues that the underlying (ὑποκείµενα) aspects of a poem, such as subject matter but also the words themselves, have authority over the poetic craft (κυρί]ων δὲ τῆς ποητικῆς ὑπαρχόντων), i.e., exercise a determinative effect over the poem and its reception: cf. Asmis (1995) 158-60. 42 Cf. On Poems 1 fr. 117.17 for φωνή as explicitly non-authorial sound. 18

comparing it to other types of sounds which exceed normative human capacities. 43 Furthermore, the thinking of the text as inscribed voice need not be privileged as a dominant way of conceiving readerly activity; rather, we need to recognize its theoretical shortcomings, as well its limitations as an historical description of what reading entailed. First, the conception of authorial φωνή is distinctly problematic as a means of thinking about performance, involving as it does a deformative construction which reduces the complex vocal and musical aspects of performance to an essentialized notion of authorial utterance. Moreover, thinking about the text as a transcription of authorial φωνή glosses over important distinctions between author and narrator, and hence limits the possibility for attention to the specific ways in which narratorial voice is constructed. 44 Authorial voice should not be conceptualized only as a productive ground which generates the text, but as a figure produced by the processes of textuality. A distinction needs to be made here between the composition and performance; in the latter, the performing voice shows up as something constructed according to generic and contingent influences, although these factors are clearly important in the compositional process as well; no author writes with a voice that is entirely his own. These issues also affect the historical resitings of texts. Even thinking in materialist terms, a later reading is a reenactment, a confluence of the reader s voice with that of the text which presupposes the absence of the actual voice of the historical author. This absence raises the question, connected to the constructedness of authorial voice just discussed, of whose voice is being reenacted the voice of Pindar the historical individual, Pindar qua constructed narrator, or the voice of the performing chorus. As my readings below will demonstrate, reading Pindaric texts often brings to the fore the impossibility of a complete reenactment of the text, not least because of the texts dramatizations of the limitations of their own enunciative strategies. 45 A voiced reading is a transformative encounter which entails the reader s physical disposition being given over to text s voice(s) while simultaneously affirming the uniquely contingent status of each recreation of the text; crucially, however, this physical dimension is only an extension of the processes of transformation implicit in any reading encounter. 43 See below pp. 85-6. 44 For an overview of the controversial issue of first persons in Pindar cf. Currie (2005) 19-21, and cf. also Fearn (2007) 7-9. The fullest treatment is Lefkowitz (1991), and note her supplementary comments in Lefkowitz (1995). A particularly interesting aspect of this problem is whether certain utterances are felt to be spoken/sung in the persona of the laudandus: Σ P. 8.78a, Σ P.9.161, Σ N.10.73b, Σ I.7.55b record judgements to this effecct. See further Currie (2005) 20-1. 45 See below pp. 187-92 and pp. 213-18. 19

The transcription-and-revivification model also ignores the considerable distantiations involved in the processes of textual dissemination and their potential to generate new meanings. Epinician s contextual aspect makes it a particularly interesting test case for addressing the operation of this dynamic. As mentioned above, strategies such as addresses to victors, deixis, and references to the environmental and musical settings of the poems prompt imaginative recreation by later readers, but they also impose a sense of the radical difference between the sites of performance and reading, a difference that would have been strengthened in later antiquity by other factors such as the oddities of Pindar s dialect, as well as the growing body of literary and scholarly receptions. When, for example, at the beginning of P.4 the narrator says that the Muse must stand today by that dear man [Arcesilas] (σάµερον µὲν χρή σε παρ ἀνδρὶ φίλῳ / στᾶµεν, 1-2) in order for the activity of praise to take place, the reader is made aware of the temporal specificity of the performance utterance, but also of his own role in continuing the text s project (αὔξῃς [sc. the Muse] οὖρον ὕµνων, P.4.3). Moreover, books are not simply transcriptions of a prior body of language but visual and tangible objects. The materialists stress on the embodied particularity of text-as-voice discounts the role of material factors such as the use of critical signs and the more general visual appearances of and physical forms of books in shaping readers access to texts. 46 An examination of Pindar s texts in later antiquity also needs to take account of interpretative and literary receptions, and in this respect the corpus of ancient scholarship represented by the Pindaric scholia is particularly important. Not only are the exegetical and historical questions foregrounded by the scholia remote from the materialist emphasis on the here-and-now of the physicality of language, 47 but they also implicate the texts they comment on in a series of intertextual, cultural, and historical networks which demand readerly attention to the contexts, both those of Pindar s own time and later, in which meaning arises and is remade, debated, and contested. In short, critical models which conceive of ancient modes of reading, silent or otherwise, either as a passive submission to the text, 48 or as a make-of-it-what-you-will exercise is postmodern relativism avant la lettre 49 should be treated with caution. I would instead conceptualize the experience of ancient reading, voiced or silent, as a dynamic process wherein the text both imposes its particular economies of meaning on its readers and is reshaped by the contexts in which it is encountered. Reading is 46 I examine this aspect of critical signs below pp. 91-104. 47 Cf. Porter (2010) 308-64. 48 The burden of Svenbro (1993) 187-216. 49 Cf. Fowler (forthcoming) 8-10. 20